Annotation of dietlibc/CHANGES, revision 1.4
1.1 cvs 1: 0.6.15:
2: clone support for i386 and alpha by Olaf.
3: Initial dynamic loading glue for i386, also by Olaf.
4: We don't have a dynamic loader yet, so this is not yet usable.
5: added readv, sigdelset, sigaltstack, sigfillset, sigismember, usleep,
6: vsprintf, writev, all by Olaf.
1.2 fefe 7: imported readdir fix from Guillaume. Mhh, can the kernel interface
8: really be this broken?
1.3 fefe 9: Found a kludgy way to detect whether the kernel includes define struct
10: stat64 (they also define STAT64_HAS_BROKEN_ST_INO). If this does
11: not work for you, please tell me!
1.4 ! fefe 12: Removed bogus __seek_types enum, it's a bunch of #defines now.
1.1 cvs 13:
14: 0.6.14:
15: *scanf did not append the 0 byte for %s.
16: added creat and changed creat to call open directly instead of open64.
17: did for fread what 0.6.10 did for fwrite.
18: repaired buffered stdio a little (line buffering is still not
19: supported).
20: [All of these bugs have been reported by Guillaume. Thanks!]
21: added __pure__ to strchr and strrchr in <string.h>
22:
23: 0.6.13:
24: included COPYING to make clear that the diet libc is covered by the
25: GNU General Public License (and _not_ the LGPL). That means that
26: you need to obtain a license from me if you want to use the diet
27: libc in a proprietary program that you want to distribute.
28: The x86 unified syscall swapped arguments #4 and #5 and noone
29: noticed until now! Thanks to Guillaume Cottenceau for reporting
30: this!
31: The malloc had a subtle bug with small allocations that could cause
32: segfault. Guillaume reported it, Olaf fixed it.
33:
34: 0.6.12:
35: added assert, statfs and fstatfs
36: added include/net/if.h so fget compiles again.
37:
38: 0.6.11:
39: added putenv.
40:
41: 0.6.10:
42: fixed fwrite. It returned the number of bytes written, not the number
43: of records. Thanks to Albert D. Cahalan for pointing this out.
44: fixed fgets. It returned EOF on empty lines. Thanks, Erik Frey.
45: added brk and sbrk. harold@nb.com.sg made me do it.
46:
47: 0.6.9:
48: fixed execvp not to return on ENOENT
49: did for {open|read|close|seek|tell}dir what I did for stat before.
50: The glibc compatibility part is currently non-functional.
51: fixed white space handling in sscanf " (" would not match " (".
52: fixed printf to accept 'l' flag.
53: fixed getpwuid (ignored last line)
54:
55: 0.6.8:
56: Olaf contributed initial thread-safe syscalls for i386, alpha, sparc,
57: mips and arm. I did the one for ppc. They can be enabled in dietfeatures.h.
58: added strlcpy and strlcat from OpenBSD.
59: added stat64, fstat64 and lstat64.
60: added endian.h and made ending of strlen.c endianness-aware
61: added string routines size tweak to dietfeatures.h. Most of the
62: string routines have been unrolled. This is up to three times as
63: fast but creates up to three times larger code. Now you can disable
64: the unrolling.
65: added sendfile.
66: I also added pread. It works on x86 and sparc but not on ppc and mips.
67: I have no idea what's going on. strace disagrees with the kernel.
68: Can anyone help? I asked the Linux kernel mailing list for help, too.
69: added a few aliases of the type __libc_open for open. nm on
70: libpthreads.so indicates that we will need them.
71:
72: 0.6.7:
73: the sources now compile without warnings with -Wall.
74: printf now prints "(null)" when %s is passed NULL. This can be
75: removed with WANT_NULL_PRINTF in dietfeatures.h
76: added vfprintf, execl, ttyname.
77: stat, lstat and fstat are now #defined to __dietstat, __dietlstat and
78: __dietfstat so they can use the normal kernel struct stat.
79: stat, lstat and fstat are now C wrappers that convert to the glibc
80: struct stat. So we can avoid that overhead for programs that use
81: dietlibc headers.
82:
83: 0.6.6:
84: changed the Makefiles so you can now set CFLAGS on the command line
85: added memccmp (analogous to memccpy) and strncmp.
86: Olaf fixed another bug in the sparc unified syscall.
87: Paul Clifford contributed a C version of his strlen.S that is much
88: more efficient than the previous strlen on all platforms! It also
89: looks like technology from Roswell. ;-}
90: added a "real" stdio. Well, almost. Please contribute!
91: added dietfeatures.h so you can remove features you don't need.
92: removed debug code from vsnprintf that changed \0 to ' ' (argh!).
93: stdio now works with simple programs and minigzip from libz.
94: stdio uses some major trickery to avoid linking the stdio, stderr and
95: stdout (and reserving space for them) when they are not used.
96:
97: 0.6.5:
98: Red Plait <redplait@ixcelerator.com> found several bugs in diet libc,
99: most of them bugs in the new header files, but also missing sigset
100: functions.
101: Paul Clifford contributed an assembly strlen.S for ARM and fixed
102: several bugs.
103: Fixed ppc/setjmp and mips/pipe.
104: Olaf contributed a new sparc unified syscall.
105:
106: 0.6.4:
107: declared is* static inline in getservent.
108: added assert_fail, strtoul, isalpha, isdigit, isalnum, isascii.
109: changed strlen to return 0 when passed NULL.
110: new, much smaller unified syscall for MIPS. Thanks to Olaf the Mad
111: Scientist who actually implemented this without access to a MIPS
112: box, just from reading the architecture manual. And it worked out
113: of the box.
114: started a set of system includes, mainly so that I can use lcc and
115: my alpha-linux cross compiler (which is unable to cross-compile
116: glibc). diet libc can now be compiled without any include files
117: from a normal libc. The includes do declare more than diet libc
118: currently delivers and they are still far from complete for real
119: applications.
120: I will only add assembly versions that are smaller _and_ faster than
121: the C version. Larger routines are only accepted if they are called
122: very often and are substantially faster.
123: added i386 assembly strchr, which is smaller and faster than the old
124: version. It is, however, larger than the version contributed by
125: proton (thanks, anyway).
126: added i386 strlen (31% faster, 14% smaller)
127: "make t" will now create a map file called "mapfile".
128:
129: 0.6.3:
130: added sys_errlist, strerror and perror
131: added isblank
132: added atol
133:
134: 0.6.2:
135: mmap for ARM didn't compile. Thanks, Paul!
136:
137: 0.6.1:
138: split mmap into the architecture specific subdirectories.
139: getenv now copes with environ==NULL, thanks Paul Clifford.
140: Paul also contributed a smaller ARM startup code.
141:
142: 0.6:
143: strcat returned the wrong result. Thanks, Dietz Pröpper.
144: strtod now understands a negative exponent (oops, thanks Bertram Barth)
145: Port to arm-linux-gnu, but on the Netwinder I use for testing the
146: __dtostr does not work (I have no idea why!)
147: The MIPS port now uses (much smaller) non-PIC code. That means that
148: the applications you link against diet libc also have to be compiled
149: non-PIC. I suggest copying the CFLAGS from the diet libc Makefile.
150: Thanks to Ralf Bächle for helping me with this!
151: I made subdirectories for the architectures and use VPATH to override
152: VPATH so that make finds the source file automatically. That should
153: simplify the sources greatly.
154: Olaf Dreesen contributed Alpha support including setjmp and longjmp!
155: ARM and MIPS now also have setjmp and longjmp
156: "compile" and "load" are now make targets. Use them for djb programs.
157: added contrib/elftrunc.c which will remove unnecessary ELF headers.
158: Again, contributed from Olaf. Great work!
159:
160: 0.5.12:
161: printf also does signed numbers.
162: If you don't use atexit, dietlibc now does not link exit, only _exit.
163: *printf now correctly returns the number of bytes written.
164: Olaf contributed experimental sscanf and vsscanf implementations.
165: If passed NULL as buffer, snprintf will not write anything but still
166: return the number of bytes it would have written.
167: Initial MIPS port! (Oh, the agony!)
168: No setjmp and longjmp support yet! Please contribute!
169: I even implemented unified syscalls for MIPS. Still, MIPS code is
170: almost twice the size of SPARC code. If anyone knows why: please
171: tell me!
172:
173: 0.5.11:
174: I implemented new unified syscalls for x86, sparc and ppc, this time based
175: on .s files and not .c files, and I moved the syscalls into
176: subdirectores to clean the dietlibc sources up a little.
177: A binary consisting of printf("%s is %d\n","olaf",23) is now
178: 2864 bytes on sparc
179: 2488 bytes on intel
180: I kludgily implemented fprintf(stdout,... and fprintf(stderr,...
181: to make a few more applications work.
182: Olaf Dreesen also implemented some unified syscall stuff for x86.
183: He discovered that with our unified syscall interface it actually
184: costs <10 bytes total to make all system calls thread safe!
185: I implemented atexit() (can register up to 4 callbacks).
186: I implemented strtod and __dtostr (the opposite). Now we can add
187: floating point support to vsnprintf and sscanf!
188:
189: 0.5.10:
190: I actually saw that I can not only merge the errno handling code of
191: the system calls on x86, I can also merge the rest (including the
192: arguments) except for setting the system call number. All those
193: system calls are now just a jump to a unified system call handler.
194: I got rid of x86openclose again. The savings are substantial:
195: chown from embutils went from 7664 to 7184 bytes!
196: If I'd move the system call wrappers to assembly language, I could
197: even reduce the alignment (does not matter since it's just a jump
198: anyway) and get rid of the "ret" behind each jump (one byte per
199: system call!)
200:
201: 0.5.9:
202: added memchr, strpbrk, strstr, strtol, isspace (hehe)
203: fixed strdup
204: fixed return values for strcpy and strcat (thanks to Norbert Berzen)
205: Olaf Dreesen contributed a strtol and initial {sn|vsn|}printf implementation
206: (no signed integers, only strings and unsigned integers (but
207: supporting octal, hex and decimal). Thanks, Olaf! (by the way:
208: that code must be wonderful, I don't understand it at all *bg*)
209: Olaf's printf does understand padding like in "%08d" and "%8d" and
210: automatically pads pointers with '0'.
211: Rewrote the x86 start code, old: 22 instructions, new: 12 instructions.
212: Thanks to proton for inspiration on this.
213: The new x86 start code also does not reference exit any more, thus saving
214: 64 bytes for executables that don't call exit explicitly.
215: [insert maniacal laughter] EVERY BYTE COUNTS!1!! ;-)
216: Moved errno and environ to start.S, saving no byte binary size but
217: speeding up compilation and slightly shortening dietlibc.a ;-)
218: For x86: joined open and close into one assembler file, sharing the
219: errno handling. This is a feasibility test and it actually saves a
220: few bytes. I think I will reimplement the _syscall[1-6] macros on
221: all platforms now to share the errno handling code for them.
222:
223: 0.5.8:
224: fixed strchr to be able to look for 0.
225:
226: added _llseek, ftruncate, getpgid, getresgid, getresuid, getsid,
227: memccpy, memmove, mprotect, setregid, setresgid, setresuid, setreuid,
228: strncpy, swapon, truncate, strtok, strtok_r, strspn, strcspn (all
229: contributed by Olaf Dreesen)
230:
231: added execvp, getcwd
232:
233: fixed __xmknod
234:
235: Note: can it be that the ftw interface really is so broken that I
236: cannot implement it without having to implement some searching data
237: structure? I included an experimental ftw implementation that will
238: not follow symlinks.
239:
240: 0.5.7:
241: oops, if_nametoindex was broken!
242:
243: 0.5.6:
244: ported to ppc-linux.
245: fixed i386 sigsetjmp (I mistyped the function name)
246: included sigjmp.c
247:
248: 0.5.5:
249: added wait, sys_siglist, longjmp/setjmp/sigsetjmp for i386 and sparc
250: actually, it wasn't fork that wasn't working for sparc, it was pipe.
251: I fixed it now.
252: dietlibc/SPARC assumes -msupersparc (does not provide div, mul, etc)
253: fixed readdir to use getdents and not the intel inline asm
254:
255: 0.5.4:
256: ported to sparclinux (sparc32 only). Beware: does not work yet.
257: added raise, abort, readlink, strcat, geteuid, geteuid, wait3, access
258:
259: 0.5.3:
260: added tcsetattr and getenv to compile e3
261:
262: 0.5.2:
263: added vhangup, tcgetattr, isatty and memcmp for fgetty
264: (http://www.fefe.de/fgetty/)
265: added localtime from uC-libc.
266:
267: 0.5.1:
268: stat and friends actually work now. Yuck! Another case of
269: translation between kernel and userland. Why can't the kernel
270: people and the libc people simply agree on a standard?
271: removed many unnecessary includes to speed up compilation.
272: added -fomit-frame-pointer and i386 compilation target to reduce code size.
273: inlined socketcall to reduce code size.
274: "load" and "compile" are examples for djb code.
275:
276: 0.5:
277: Olaf Dreesen contributed a much smaller implementation of malloc and friends.
278: each object file is now treated with "strip -x -R .note -R .comment"
279: split each system call into a separate object file
280: added ntohs, htons, alarm, if_indextoname and if_nametoindex
281:
282: 0.4.1:
283: oops, the strchr implementation was wrong. Thanks Jens Laas!
284:
285: 0.4:
286: copied opendir and friends from uC-libc.
287: added getservent and getservby* in gerservent.c
288:
289: 0.3:
290: remove readdir system call
291: added getdents system call
292: added getpwnam/getpwuid implementation in getpwnam.c
293: added a few string functions (in str*.c)
294:
LinuxTV legacy CVS <linuxtv.org/cvs>